Test Setup and methology.

*AMD64 FX55
*OCZ PC-4000 2GB (2x1024MB) gold edition dual channel kit
*OCZ PC4000 1GB (2 x 512MB) dual channel gold series
*Thermalright XP120 with YSTECH 120mm fan
*Artic Silver 5
*DFI Ultra-D Nforce4 motherboard
*XFX 7800GTX 256meg PCIe video card
*300gig Maxtor 16 meg cache SATA HD
*Sony 16x DVD Burner
*Audigy 2 ZS
*OCZ 600w Powerstream PSU
*Windows XP Professional SP2

Finding the maximum (stable) overclock takes quite some time and plenty of patience, armed with tools such as memtest and Prime95 (and experimenting with various timings) testing was carried out over the course of a week. Using the FX makes this somewhat easier as we have the full range of overclocking options at hand, prior to this test I was already aware of the maximum stable overclock of the processor on air (2.82ghz) so we could work within these boundaries in finding the maximum memory speeds without worrying about the cpu flagging errors. I found the maximum stable overclocks with this memory to be 276x10 running at 1:1. The timings of 3-4-4-8 we found in this case the optimum rating, anything tighter at high speeds resulted in errors regardless of voltage.

OCZ guru Tony Leach started an interesting thread over on bleedinedge forums and I agree with his findings, these modules do not work better on any voltage over 2.7/2.8, Tony has also mentioned they are not being marketed for extreme overclockers. At 2.7 I found the "sweet" spot for the maximum figures mentioned earlier, some are finding more success with 2.8volts however that was not my experience. It is also worth mentioning that having tested some 1 gigabyte sticks many months ago I was pleased to see these OCZ modules can handle 1T timings at high frequencies, moving from 2T to 1T is a massive real world performance boost and an important point many of you will already be wondering about. To check total stabilty, memtest was left to run overnight for a period of 10-12 hours.

"Is 2.5 CAS possible?" the quick and simple answer is "no". With 1024MB on each stick it looks like CAS2.5 or CAS2 are only possibilites for the future and will require some serious engineering wizardry.

For comparison purposes in the synthetic tests ive included some results from a prior OCZ EL DDR PC-5000 Platinum (DFI Nf4 special) review, Ideally I would have liked to rerun the tests with the FX55 however I no longer have the OCZ PC-5000 Platinum modules. It however is still a worthwhile comparison.

In regards to the game tests its important to try and keep the timings as close as possible to ascertain the benefits of simply adding more memory of identical performance, therefore i've included OCZ PC4000 1GB (2 x 512MB) dual channel gold series CAS 2.5 results, I loosened CAS from 2.5 to 3 to match the OCZ 2GIG kit. Why? well to measure any benefits by simply adding another 1 gig of system memory. Obviously (but i'll spell it out), this is not a direct comparison of OCZ 1GIG and 2GIG DDR4000 products (as timings have been altered to an equal footing).... but more an overall approximation of game performance.

 

Previous: Introduction

Next: Synthetic Testing